Professional Documents
Culture Documents
http://robots.stanford.edu/cs223b/index.html 1
* Pictures from Mean Shift: A Robust Approach toward Feature Space Analysis, by D. Comaniciu and P. Meer http://www.caip.rutgers.edu/~comanici/MSPAMI/msPamiResults.htm
Outline
• Segmentation Intro
– What and why
– Biological
Segmentation:
• By learning the background
• By energy minimization
– Normalized Cuts
• By clustering
– Mean Shift (perhaps the best technique to date)
• By fitting
– optional, but projects doing SFM should read.
Reading source: Forsyth Chapters in segmentation, available (at least this term)
http://www.cs.berkeley.edu/~daf/new-seg.pdf
2
Intro: Segmentation and Grouping
What: Segmentation breaks an image into groups over space and/or time
Why:
4
Biological:
For humans at least, Gestalt psychology identifies several properties that result
In grouping/segmentation:
5
Biological:
For humans at least, Gestalt psychology identifies several properties that result
In grouping/segmentation:
6
Consequence:
Groupings by Invisible Completions
7
* Images from Steve Lehar’s Gestalt papers: http://cns-alumni.bu.edu/pub/slehar/Lehar.html
Consequence:
Groupings by Invisible Completions
8
* Images from Steve Lehar’s Gestalt papers: http://cns-alumni.bu.edu/pub/slehar/Lehar.html
Consequence:
Groupings by Invisible Completions
9
* Images from Steve Lehar’s Gestalt papers: http://cns-alumni.bu.edu/pub/slehar/Lehar.html
Here, the 3D nature of grouping is apparent:
Why do these tokens belong together?
In
The (in) line at the far
end of corridor must
be longer than the (out)
near line if they measure
to be the same size
Out
10
And the famous invisible dog eating
under a tree:
11
Background Subtraction
12
Background Subtraction
1. Learn model of the background
– By statistics (µ ,σ ); mixture of Gaussians; Adaptive filter, etc
1. Take absolute difference with current frame
– Pixels greater than a threshold are candidate foreground
1. Use morphological open operation to clean up point
noise.
2. Traverse the image and use flood fill to measure size of
candidate regions.
– Assign as foreground those regions bigger than a set value.
– Zero out regions that are too small.
1. Track 3 temporal modes:
(1) Quick regional changes are foreground (people, moving cars);
(2) Changes that stopped a medium time ago are candidate
background (chairs that got moved etc);
(3) Long term statistically stable regions are background.
13
Background Subtraction Example
14
Background Subtraction Principles
At ICCV 1999, MS Research presented a study, Wallflower: Principles and Practice
of Background Maintenance, by Kentaro Toyama, John Krumm, Barry Brumitt, Brian
Meyers. This paper compared many different background subtraction techniques
and came up with some principles:
P1:
P2:
P3:
P4:
P5: 15
Background Techniques Compared
16
F
Segmentation by Energy
Minimization:
Graph Cuts
17
Graph theoretic clustering
• Represent tokens (which are associated with
each pixel) using a weighted graph.
– affinity matrix (pi same as pj => affinity of 1)
• Cut up this graph to get subgraphs with strong
interior links and weaker exterior links
19
* From Khurram Hassan-Shafique CAP5415 Computer Vision 2003
Weighted Graphs and Their
Representations
0 1 3 ∞ ∞
a 1 0 4
b ∞ 2
3 4 0 6 7
c
e ∞ ∞ 6 0 1
6 ∞ 2 7 1 0
d Weight Matrix: W
20
* From Khurram Hassan-Shafique CAP5415 Computer Vision 2003
Minimum Cut
A cut of a graph G is the set of
edges S such that removal of
S from G disconnects G.
21
* From Khurram Hassan-Shafique CAP5415 Computer Vision 2003
Minimum Cut and Clustering
22
* From Khurram Hassan-Shafique CAP5415 Computer Vision 2003
Image Segmentation & Minimum Cut
Pixel
Neighborhood
w
Image
Pixels
Similarity
Measure
Minimum
Cut
23
* From Khurram Hassan-Shafique CAP5415 Computer Vision 2003
Minimum Cut
• There can be more than one minimum cut in a
given graph
1
H. Nagamochi, K. Nishimura and T. Ibaraki, “Computing all small cuts in an
undirected network. SIAM J. Discrete Math. 10 (1997) 469-481.
24
* From Khurram Hassan-Shafique CAP5415 Computer Vision 2003
Finding the Minimal Cuts:
Spectral Clustering Overview
25
* Slides from Dan Klein, Sep Kamvar, Chris Manning, Natural Language Group Stanford University
Eigenvectors and Blocks
• Block matrices have block eigenvectors:
λ 1= 2 λ 2= 2 λ 3= 0 λ 4= 0
1 1 0 0 .71 0
1 1 0 0 .71 0
0 0 1 1 eigensolver 0 .71
0 0 1 1 0 .71
• Near-block matrices have near-block eigenvectors: [Ng et
al., NIPS 02]
λ 1= 2.02 λ 2= 2.02 λ 3= -0.02 λ 4= -0.02
1 1 .2 0 .71 0
1 1 0 -.2 .69 -.14
.2 0 1 1 eigensolver .14 .69
0 -.2 1 1 0 .71
26
* Slides from Dan Klein, Sep Kamvar, Chris Manning, Natural Language Group Stanford University
Spectral Space
• Can put items into blocks by eigenvectors: e1
1 1 .2 0 .71 0
1 1 0 -.2 .69 -.14
.2 0 1 1 .14 .69 e2
0 -.2 1 1 0 .71
e1 e2
• Clusters clear regardless of row ordering:
e1
1 .2 1 0 .71 0
.2 1 0 1 .14 .69
1 0 1 -.2 .69 -.14 e2
0 1 -.2 1 0 .71
e1 e2
27
* Slides from Dan Klein, Sep Kamvar, Chris Manning, Natural Language Group Stanford University
The Spectral Advantage
• The key advantage of spectral clustering is the
spectral space representation:
28
* Slides from Dan Klein, Sep Kamvar, Chris Manning, Natural Language Group Stanford University
Clustering and Classification
• Once our data is in spectral space:
– Clustering
– Classification
29
* Slides from Dan Klein, Sep Kamvar, Chris Manning, Natural Language Group Stanford University
Measuring Affinity
Intensity
f()
ax
,y=
e
x
p
1
i
2
σ
(
− 2I)
x−
I(
y(
)
2
)
Distance
f()
ax
,y=
e
xp
−
1
2σ
2
d
( )
x−
y
2
Texture
f()
ax
,y=
e
x
p
−
1
t
2
σ
2((
c
x)
−(
c)
2
y
)
30
* From Marc Pollefeys COMP 256 2003
Scale affects affinity
31
* From Marc Pollefeys COMP 256 2003
32
* From Marc Pollefeys COMP 256 2003
Drawbacks of Minimum Cut
• Weight of cut is directly proportional to the
number of edges in the cut.
Cuts with
lesser weight
than the
ideal cut
Ideal Cut
33
* Slide from Khurram Hassan-Shafique CAP5415 Computer Vision 2003
Normalized Cuts 1
w( A, B ) w( A, B )
N cut ( A, B ) = +
∑ x∈ A, y∈V
w( x, y ) ∑ z∈B , y∈V
w( z , y )
Aug 2000. 34
* Slide from Khurram Hassan-Shafique CAP5415 Computer Vision 2003
Finding Minimum Normalized-Cut
• Finding the Minimum Normalized-Cut is
NP-Hard.
• Polynomial Approximations are generally
used for segmentation
35
* Slide from Khurram Hassan-Shafique CAP5415 Computer Vision 2003
Finding Minimum Normalized-Cut
e − Fi − F j σ F2
×e
− X i − X j σ X2
if j ∈ N ( i )
W ( i, j ) =
0 otherwise
36
* From Khurram Hassan-Shafique CAP5415 Computer Vision 2003
Finding Minimum Normalized-Cut
yT ( D − W) y
• It can be shown that min N cut = min y
y T Dy
37
* Slide from Khurram Hassan-Shafique CAP5415 Computer Vision 2003
Algorithm
• Compute matrices W & D
• Solve ( D − W ) y = λDy for eigen vectors with the
smallest eigen values
• Use the eigen vector with second smallest eigen value
to bipartition the graph
• Recursively partition the segmented parts if necessary.
38
* Slide from Khurram Hassan-Shafique CAP5415 Computer Vision 2003
Figure from “Image and video segmentation: the normalised cut framework”,
by Shi and Malik, 1998
39
* Slide from Khurram Hassan-Shafique CAP5415 Computer Vision 2003
F igure from “Normalized cuts and image segmentation,” Shi and Malik, 2000
40
* Slide from Khurram Hassan-Shafique CAP5415 Computer Vision 2003
Drawbacks of Minimum Normalized Cut
41
* Slide from Khurram Hassan-Shafique CAP5415 Computer Vision 2003
Segmentation by Clustering
42
Segmentation as clustering
• Cluster together (pixels, • Point-Cluster distance
tokens, etc.) that belong – single-link clustering
together – complete-link clustering
• Agglomerative clustering – group-average clustering
– attach closest to cluster it • Dendrograms
is closest to – yield a picture of output as
– repeat clustering process
continues
• Divisive clustering
– split cluster along best
boundary
– repeat
43
* From Marc Pollefeys COMP 256 2003
Simple clustering algorithms
44
* From Marc Pollefeys COMP 256 2003
45
* From Marc Pollefeys COMP 256 2003
Mean Shift Segmentation
• Perhaps the best technique to date…
http://www.caip.rutgers.edu/~comanici/MSPAMI/msPamiResults.html
46
Mean Shift Algorithm
Mean Shift Algorithm
1. Choose a search window size.
2. Choose the initial location of the search window.
3. Compute the mean location (centroid of the data) in the search window.
4. Center the search window at the mean location computed in Step 3.
5. Repeat Steps 3 and 4 until convergence.
The mean shift algorithm seeks the “mode” or point of highest density of a data distribution:
47
Mean Shift Segmentation
Mean Shift Setmentation Algorithm
1. Convert the image into tokens (via color, gradients, texture measures etc).
2. Choose initial search window locations uniformly in the data.
3. Compute the mean shift window location for each initial position.
4. Merge windows that end up on the same “peak” or mode.
5. The data these merged windows traversed are clustered together.
48
*Image From: Dorin Comaniciu and Peter Meer, Distribution Free Decomposition of Multivariate
Data, Pattern Analysis & Applications (1999)2:22–30
Mean Shift Segmentation Extension
Is scale (search window size) sensitive. Solution, use all scales:
Gary Bradski’s internally published agglomerative clustering extension:
Mean shift dendrograms
1. Place a tiny mean shift window over each data point
2. Grow the window and mean shift it
3. Track windows that merge along with the data they transversed
4. Until everything is merged into one cluster
Best 4 clusters: Best 2 clusters:
49
Advantage over agglomerative clustering: Highly parallelizable
Mean Shift Segmentation
Results:
50
http://www.caip.rutgers.edu/~comanici/MSPAMI/msPamiResults.html
K-Means
• Choose a fixed number • Algorithm
of clusters – fix cluster centers;
allocate points to
• Choose cluster centers closest cluster
and point-cluster – fix allocation; compute
allocations to minimize best cluster centers
error •
x could be any set of
• can’t do this by search, features for which we
because there are too can compute a
many possible distance (careful about
allocations. scaling)
∑ ∑x µ
2
j− i
∈
icl
u
s
t
es
r ∈
jel
e
m
en
t
so
f
i
' t
h
c
lu
s
t
er 51
* From Marc Pollefeys COMP 256 2003
K-Means
52
* From Marc Pollefeys COMP 256 2003
Image Segmentation by K-Means
• Select a value of K
• Select a feature vector for every pixel (color, texture,
position, or combination of these etc.)
• Define a similarity measure between feature vectors
(Usually Euclidean Distance).
• Apply K-Means Algorithm.
• Apply Connected Components Algorithm.
• Merge any components of size less than some
threshold to an adjacent component that is most
similar to it.
53
* From Marc Pollefeys COMP 256 2003
Results of K-Means Clustering:
55
RANSAC
• Choose a small subset • Issues
uniformly at random – How many times?
• Fit to that • Often enough that we are
likely to have a good line
• Anything that is close to – How big a subset?
result is signal; all others • Smallest possible
are noise – What does close mean?
• Refit • Depends on the problem
56
* From Marc Pollefeys COMP 256 2003
57
* From Marc Pollefeys COMP 256 2003
Distance threshold
Choose t so probability for inlier is α (e.g. 0.95)
• Often empirically
2
• Zero-mean Gaussian noise σ then d ⊥ follows
χ m2 distribution with m=codimension of model
(dimension+codimension=dimension space)
Codimension Model t2
1 line,F 3.84σ2
2 H,P 5.99σ2
3 T 7.81σ2
58
* From Marc Pollefeys COMP 256 2003
How many samples?
Choose N so that, with probability p, at least one
random sample is free from outliers. e.g. p=0.99
(1 − (1 − e) )
s N
= 1− p
(
N = log(1 − p ) / log 1 − (1 − e )
s
)
proportion of outliers e
s 5% 10% 20% 25% 30% 40% 50%
2 2 3 5 6 7 11 17
3 3 4 7 9 11 19 35
4 3 5 9 13 17 34 72
5 4 6 12 17 26 57 146
6 4 7 16 24 37 97 293
7 4 8 20 33 54 163 588
59
8 5 9 26 44 78 272 1177
* From Marc Pollefeys COMP 256 2003
Acceptable consensus set?
• Typically, terminate when inlier ratio reaches
expected ratio of inliers
T = (1 − e ) n
60
* From Marc Pollefeys COMP 256 2003
Adaptively determining the
number of samples
e is often unknown a priori, so pick worst case, e.g. 50%, and
adapt if more inliers are found, e.g. 80% would yield e=0.2
– N=∞, sample_count =0
– While N >sample_count repeat
• Choose a sample and count the number of inliers
• Set e=1-(number of inliers)/(total number of points)
• Recompute N from e
• Increment the sample_count by 1
– Terminate
( N = log(1 − p ) / log(1 − (1 − e) )) s
61
* From Marc Pollefeys COMP 256 2003
RANSAC for Fundamental Matrix
Γ = 1 − (1 − ( # matches
)
# inliers 7 # samples
)
}
3.3.1 verify if inlier
while hypothesis is still promising (verify hypothesis)
while Γ (#inliers,#samples)<95%
63
* From Marc Pollefeys COMP 256 2003
Example: robust computation
from H&Z
Interest points
(500/image)
(640x480)
#in 1-e adapt. N
6 2% 20M
10 3% 2.5M
44 16% 6,922
Putative
58 21% 2,291
correspondences (268)
73 26% 911
151 56% 43
(Best match,SSD<20,±320)
Outliers (117)
(t=1.25 pixel; 43 iterations)
Inliers (151)
• Enhancements to RANSAC
– Randomized RANSAC
– Sample ‘good’ matches more frequently
– …
65
* From Marc Pollefeys COMP 256 2003